RISC-V has been an industry buzzword over the last few years, making waves with a range of wacky devices and chips from all sorts of manufacturers. This often-hyped technology has sometimes struggled to live up to expectations, but it is slowly looking more poised to revolutionize the semiconductor market.

But what makes RISC-V so exciting, and why does everyone from NVIDIA to Microsoft seem to be investing in it?

What is RISC-V?

An alternative instruction set architecture👁 RISC-V logo

RISC-V is an instruction set architecture (ISA). An ISA is a template for the set of instructions a CPU must be able to interpret and handle, as well as specifications for how they should function. You can think of it as a programming language for CPUs. ISAs are important because the ISA is what makes software portable from one CPU to the next - it provides guarantees that all the individual instructions of your software are supported and available on the CPU, and that they function in a specific way. RISC-V (pronounced Risk-five) was developed by academics at Berkeley's Parallel Computing Lab in 2010 and is now managed by RISC-V International. The first RISC chips were in production in 2011, and since then they have received funding from everyone from Microsoft to nation states.

RISC-V is not inherently faster or even better than Arm or Intel's offerings, but as we'll see, that might not matter.

RISC-V is a free alternative ISA

This open standard requires no expensive licensing fees👁 Screenshot of the RISC-V Github repo.

One key commercial difference between RISC-V and other common ISAs is the fact that RISC-V is an open-source license. This means that companies are free to use it, and profit from it, without a need to pay licensing fees to the owners of the license. Those types of fees can often be significant, and get indirectly passed on to both consumers and businesses as part of the cost of manufacturing products. It's common for Arm and Intel to also charge fees for modifying their ISAs, making significant divergence even more expensive.

RISC-V is helping to resolve a steep hurdle for hardware companies, since many companies may want to develop their own ISA to save on the licensing fees, but only a few are operating on the scale required to have even a chance of being profitable (if they can even succeed in matching standard performance and features). This is why many big companies, including Google, Nvidia, Red Hat, Samsung, and even NVIDIA, are contributing significant engineering resources to the development and maturity of RISC-V.

Easier hardware design is helping to fuel a custom hardware boom

RISC-V is quietly enabling a divergence in custom hardware for domain-specific applications, by providing an easy (or at least easier) pathway for businesses and academics to build their own versions of hardware when off-the-shelf options aren't suitable. This works in tandem with the wide range of full open-source RISC-V implementations on the market. Businesses may be able to take an existing open-source implementation of RISC-V (effectively a design for a complete processor core, usually written in a dedicated language like Verilog/SystemVerilog) and make modifications to suit their specific use case. This can involve dropping aspects that aren't needed, and adding pre-bundled supporting elements to the core which may even be off-the-shelf elements. This means that where previously it wouldn't have been practical or affordable to build specific hardware for a feature, it's now more broadly possible.

Companies small and large are already utilizing this technology. It's difficult to understand if companies are designing cores from the ground up or using pre-constructed designs, but custom RISC-V silicon has already made its way into the market. For example, Google launched their Titan M2 security modules with the Pixel 6, and we recently saw a groundbreaking RISC-V SOC combining a GPU and CPU.

RISC-V is driving innovation

A lower barrier to entry makes innovation cheaper and faster

Given the licensing of RISC-V, there's nothing to stop companies from taking an existing open-source RISC-V design, iterating and improving on it, then selling it back to the market (often with a similar licensing model to Arm) for a premium... and this is already happening. The accessibility of RISC-V allows these companies to both come in at a lower unit cost (without licensing fees due to Arm or Intel) focus on improving these open-source designs to optimize them for their specific use case. This is building a whole ecosystem of companies competing against each other for efficient RISC-V based chips. This kind of market does exist for Arm/x86, but is crippled by the modification fees and high barrier to entry imposed by licensing fees, which restrict its access to the mainstream players (i.e. Qualcomm).

Similarly, as more hardware becomes available (both specialized and generalized), more developer support accrues, adding more software availability, support, and knowledge. This is what's known as a virtuous circle, where interest, innovation, and investment amplify to create a positive feedback loop.

This has been achieved by lowering the barrier of entry to the market, making it cheaper to develop new products and ultimately usher in wider innovation. If it continues, we could quickly see an accelerating pace of increasingly commetetive RISC-V chips.

RISC-V is challenging monopolies

Could ARM and Intel's days at the top be numbered?

Despite being barely a decade old, RISC-V is proving to be successful in challenging the monopolies within chip manufacturing.

CPU production is a complex market, with an extremely high barrier to entry due to significant design, validation, and verification required before we even start talking about performance optimization. Engineering talent is expensive and difficult to find, so many companies have been skipping custom hardware altogether in favor of a faster time to market and lower upfront cost. Fundamentally, it's this economic calculus that has driven the monopoly of x86 and ARM for so long; but RISC-V is changing that. By providing a cheaper option to bring either sold as-is custom chips to market, or products containing custom chips, companies are more likely to see economic benefits sooner with the lower upfront cost it enables. As more companies take up this opportunity, the more appealing it becomes.

This is how RISC-V is slowly and quietly changing the CPU market; not by being an intrinsically faster or more efficient standard than offered by ARM and Intel, but by driving innovation enough to allow it to compete.

RISC-V has the resources to catch up with the industry leaders

This is where the real excitement comes from with RISC-V. As the industry for RISC-V starts to reach that critical mass, and the research, investment, and innovation continue to push it forward, the combined resources working to develop faster and more efficient RISC-V chips will far outpace that of the other major players in the CPU space. This is what might ultimately lead RISC-V to be a superior and even cheaper product, since companies won't be doing all the work themselves. You might envisage a world where companies are able to take hyperefficient open-source designs, make small optimizations, and sell them back to the market at a significantly lower cost than what a private company would ever be able to achieve with a completely closed-source proprietary design. This is part of what makes RISC-V both important and exciting to the industry and our future.